| Quality inspection management refers to measures to control various procedures and the quality of finished products in the production and processing process.It is an important link to control product quality and a hot issue for sales companies.Traditional quality inspection management focuses on product quality inspection but ignores the problem of automatic product classification.With the deepening of research,the quality inspection management method based on machine learning has solved the product classification problem well.But there are still some deficiencies.On the one hand,the quality inspection management method based on machine learning still has a lot of room for improvement in the category imbalance problem;on the other hand,the existing quality inspection management method based on machine learning does not consider the issue of supplier quality prediction running time.In order to solve the above two problems,two methods of ACS-IGBDT and PSO-MLRM are proposed respectively.First,in order to solve the problem of unbalanced product classification sample in the quality inspection management method,this paper applies the improved gradient lifting iterative decision tree IGBDT to quality inspection management,Through cost-sensitive learning,the idea of introducing different classification result costs during model training is adopted to achieve the effect of solving the problem of class imbalance.A stochastic gradient lifting algorithm IGBDT based on cost-sensitive learning is proposed.The experimental results show that the accuracy and precision of the IGBDT method is higher than the existing mainstream classification methods in the quality inspection management problem.Secondly,in order to solve the problem that the newly added products in quality inspection management fail to recommend the best quality inspection program,this paper combines the concept of cosine similarity optimization on the basis of IGBDT,and proposes an improved decision tree quality based on cosine similarity optimization The inspection plan recommends the ACS-IGBDT model,and an experimental comparison is made on the wine data set of product classification results.The results show that ACSIGBDT has certain practicability and superiority in the field of quality inspection.Finally,in order to shorten the running time of supplier quality prediction,this paper proposes a PSO-MLRM model based on particle swarm multiple linear regression quality prediction.According to the idea of particle swarm optimization algorithm,the model uses each set of observation data as the particles of the algorithm,and iteratively finds the optimal solution of the model group with the smallest residual error.The model mainly reduces the running time by reducing the parameters and the convergence speed.Finally,this paper uses the PSO-MLRM model to compare and verify the data set downloaded on the UCI data warehouse.The results show that PSO-MLRM has better performance in supplier quality prediction.In summary,the paper designed and developed a quality inspection management system,and used the recent product quality inspection data to empirically demonstrate the PSO-MLRM and ACS-IGBDT models mentioned above in the system.The results prove that the model is in the task of product quality inspection management Both accuracy and robustness have achieved the expected results,and are put into the application of quality inspection management in actual enterprises. |
